Files
msd-core/tests
Tom Boucher ae7dc52972 fix(#3225): guard W006/W007 + consistency loops with isSentinelPhaseId (#3371)
* test(#3225): sentinel phase dirs no longer trigger W007 / consistency warnings / gaps

The W006/W007 (validate health) and the parallel consistency disk↔roadmap and
gap-numbering loops never got the isSentinelPhaseId guard that phase.cts has
(#2786/#2949), so every sentinel phase dir (999.x/0.x — never-on-roadmap by
convention) produced a spurious W007 and a spurious 'Gap in phase numbering:
N → 999'. Add failing-first regressions for both surfaces + the gap check, each
with a non-sentinel orphan negative-space guard.

RED — fails on next; fix follows.

* fix(#3225): guard W006/W007 + consistency + gap loops with isSentinelPhaseId

cmdValidateHealth's W006/W007 loops, cmdValidateConsistency's parallel disk↔
roadmap loops, AND its gap-numbering check never got the isSentinelPhaseId guard
that phase.cts has at 10+ sites (#2786/#2949). So any repo using the sentinel-id
convention (999.x backlog/interim, 0.x drafts) got a permanent spurious W007 and
a spurious 'Gap in phase numbering: N → 999', with advice to add-to-roadmap
(violates the convention) or delete (destroys archived work).

Add isSentinelPhaseId to the phaseIdMod destructure and skip sentinel ids in:
W006 + W007 (cmdValidateHealth); the two plain-warning disk↔roadmap loops and
the gap-numbering integerPhases filter (cmdValidateConsistency — same bug family,
folded in inline per no-silent-defer). Additive only: non-sentinel orphans and
real numbering gaps still warn. The gap-numbering guard was surfaced by the
isolated review (a 999-interim dir would otherwise create a false 'N → 999' gap).
Same family as #3167 (since fixed).

* chore(#3225): add changeset fragment

* chore(#3225): backfill changeset PR number (#3371)

---------

Co-authored-by: sim <sim@local>
2026-08-11 21:49:39 -04:00
..